The likelihood of transfusion of transmissible infection via blood components and fractionated blood products has declined dramatically in the last two decades. In the early 1970s, the risk of hepatitis transmission to recipients of blood or blood components was between 10% and 25% (Horowitz & Ben-Hur, 1995). Today, there is still a residual risk associated with labile blood components [fresh frozen plasma (FFP), platelets and red cells]. Recently, in the UK, this has been estimated for human immunodeficiency virus (HIV)-1 and -2 as 0·19/million donations, for hepatitis C virus (HCV) as 0·6/million donations and for hepatitis B virus (HBV) as 5–20/million donations (L. M. Williamson, personal communication). This risk reduction has been achieved largely by improved donor selection procedures and the screening of donated blood for agents that may cause transfusion-transmitted infections (TTI). Pooling of plasma donations to make fractionated blood products such as coagulation factor concentrates and intravenous immunoglobulin (IVIG) increases the chance of viral transmission. In the USA, between 1979 and 1985, 70% of severe haemophiliacs became infected with HIV-1 that contaminated factor VIII concentrates (Goedert et al, 1989). Since then, a number of viral inactivation strategies including dry heat, heating solution and solvent–detergent (SD) treatment have been developed for fractionated plasma products. These have resulted in an enormous reduction in risk such that the likelihood of viral transmission from such fractionated blood components can now only be estimated by calculations of probability. These are based on factors such as the risk per donor, quantity of virus that may be present in a unit that gives a negative test, the number of units pooled and estimations of virus killing and removal during the process. In 1990, this was determined for a factor VIII concentrate purified by a monoclonal antibody affinity method and SD treated. It was calculated that the risk for an HIV-1, HBV or HCV transmission was < 1 in 1016, in 1013 and in 106 vials administered respectively (Horowitz & Ben-Hur, 1995). This represented, at the time, a risk reduction of between 1000 and 10 billion compared with untreated single unit blood components. After licensing for the treatment of coagulation factor concentrates in 1985, it was shown at the New York Blood Center that SD treatment could be used for the viral inactivation of FFP (Horowitz et al, 1992). In 1991, it was estimated that > 3 million doses of SD-treated FFP prepared from pooled plasma had been administered in Europe (Horowitz et al, 1998). This process was licensed by the Food and Drug Administration (FDA) in 1999. During this time, it was also shown that the use of methylene blue (MB) in conjunction with visible light had significant antiviral activity (Lambrecht et al, 1991; Bachmann et al, 1995). In 1992, MB-FFP was introduced into clinical use by several Red Cross Transfusion Services in Germany and Switzerland (Mohr et al, 1997). In the 1990s, there were a number of reports describing the photodynamic treatment with psoralens and ultraviolet A (UVA) light to inactivate viruses in platelet concentrates (Dodd et al, 1991; Margolis-Nunno et al, 1997; Grass et al, 1998). Recently, viruses and other pathogens have been inactivated in FFP using a combination of the psoralen S-59 and UVA light (Grass et al, 1998). It is important to balance the potential benefits of reduced viral transmission against the cost and logistics of implementing viral inactivation strategies, although in theory it is desirable that the risk of TTI is minimized wherever possible. Therefore, there is a reluctance to ignore new strategies that may help to achieve this and a need to assess their potential contribution to blood safety. In this article, the use of SD treatment and the use of MB plus white light and the psoralen S-59 plus UVA light are discussed and progress made towards their clinical implementation is described. In the UK, this is based on rigorous donor selection criteria to eliminate those who may be at particular risk of transmitting infection. In addition, FFP is not made from first-time donors. This is coupled with testing of all donations for HIV-1 and -2 (HIV-1 and -2 antibodies), HCV (HCV antibodies) and HBV (HBsAg) (Guidelines for Blood Transfusion Services in the United Kingdom, 2000). Blood Services in the UK also introduced PCR testing for HCV-RNA in 1999 for FFP but not cellular products. In addition, in some countries, blood centres routinely use additional tests, including alanine aminotransferase, anti-HBc, PCR for HCV-RNA, p24 antigen (HIV-1) and HTLV-1 antibody. Plasma pools used to make solvent–detergent-treated FFP may also be tested by PCR for hepatitis A (HAV), HBV, HCV, HIV-1 and parvovirus B-19. However, in the UK, blood is not tested for HAV and parvovirus B-19. For fractionated plasma products, the risk is reduced further by partitioning of viruses during the fractionation process before viral inactivation. There is good evidence that HTLV-1 and cytomegalovirus (CMV) are not transmitted by FFP (Bowden & Sayers, 1990; Donegan et al, 1994), but there may be a number of other infectious agents, potentially transmissible via labile blood components, which may not be identified or known at the present time to be pathogenic. It is important not to increase the chance of TTI by transfusing blood components unnecessarily and there is ample evidence that for FFP this is often the case. Guidelines produced by the National Institutes for Health (NIH) (Consensus Conference, 1985) and the British Committee for Standards in Haematology (BCSH) (Contreras et al, 1992) indicate where it is appropriate to transfuse FFP (Table I). The BCSH indications subdivide indications for transfusion of FFP into definite and conditional. Both guidelines indicate that FFP should not be used for nutritional support, volume expansion and as ‘formula replacement’ in patients with massive haemorrhage, e.g. administration of one unit of FFP for each 4–6 units of blood transfused. In addition, the BCSH guidelines do not recommend transfusion of FFP as a source of immunoglobulin in inherited immunodeficiency states (Contreras et al, 1992). Despite this guidance, there is evidence from audits that FFP is often transfused inappropriately (Snyder et al, 1986; Thomson et al, 1991; Schots & Steenssens, 1994). Lack of adherence to clinical guidelines was illustrated in the European Sanguis study, in which the use of FFP in total hip replacement, coronary artery bypass grafting and abdominal aortic aneurysm surgery varied from < 5% to > 95% of cases (Sirchia et al, 1994). In this study, the most common reason for receiving a transfusion of any sort was most specifically related to the hospital in which the operation was undertaken and, within that, to the hospital team performing surgery (Sirchia et al, 1994). Where clinicians completed a questionnaire specifying their triggers for prescribing FFP, the commonest indications were bleeding (43%), abnormal coagulation tests (26%) and signs/symptoms of hypovolaemia (16%) (Snyder et al, 1986). Our understanding of techniques for inactivating viruses in blood products derives in large part from the fractionation of plasma to make albumin, intravenous immunoglobulin (IVIG) and coagulation factor concentrates. The physical separation of plasma proteins is achieved primarily by adjustment of pH, temperature and ethanol concentration (Knowles, 1995). This process is known as Cohn cold ethanol extraction. Further purification steps include ion exchange, affinity and size exclusion chromatography and polyethyleneglycol precipitation. Subsequently, a number of techniques including dry heat, heating in solution, solvent–detergent treatment and β1 propiolactone/ultraviolet light treatment may be used. Most of these steps result in a significant reduction of infectious viruses. Cold ethanol fractionation itself removes 5 log of HIV (Burnouf-Radosevich et al, 1992) and other viruses, including pseudorabies, bovine viral diarrhoea virus (BVDV), Sindbis, vaccinia and vesicular stomatitis viruses (VSV) (Chandra et al, 1999) but not hepatitis viruses (Burnouf-Radosevich et al, 1992). Affinity chromatography is active in removing both enveloped (Sindbis) and non-enveloped (polio) viruses (Roberts et al, 1994). Pasteurization (60°C for 10 h) kills a wide range of both enveloped and non-enveloped viruses (Burnouf-Radosevich et al, 1992; Chandra et al, 1999). Combined treatment with β-propiolactone, detergent and ultraviolet (UV) C light was shown to inactivate 6·9 log of hepatitis B and non-A, non-B viruses in a chimp model and is used in the manufacture of some factor (F)IX concentrates (Prince et al, 1983). In other studies, UVB or UVC irradiation, laser-pulsed UVB irradiation and gamma irradiation (2·5–10 MRad) were found to inactivate between 4 and 6 log of polio, vaccinia, polio and HSV-1 and HIV-1 (Prodouz et al, 1987; Hiemstra et al, 1991; Hart et al, 1993). Solvent–detergent (SD) treatment is highly effective in killing a number of viruses but is relatively ineffective against non-enveloped viruses (Mitra et al, 1994). A number of chemicals damage viruses and have been evaluated for viral inactivation. These include sodium hypochlorite and dichloroisocyanurate (duck HBV) (Tsiquaye & Barnard, 1993) and members of the imine family such as N-acetylethyleneimine (polio and foot and mouth disease viruses). Methylene blue (MB) has good activity against enveloped viruses (Brown et al, 1998). Three principal technologies have emerged for the virus inactivation (VI) of unfractionated plasma. These are SD, MB plus white light treatment and the use of a novel psoralen (S-59) followed by UVA exposure. Despite the use of viral inactivation procedures, there was concern after outbreaks of HAV infection occurred in patients with haemophilia A who were treated with SD-treated FVIII concentrates in Italy, Belgium, Germany and Ireland (Gerritzen et al, 1992; Mannucci, 1992; Temperley et al, 1992; Peerlinck & Vermylen, 1993). The cases that were described occurred between 1988 and 1993 and were characterized by jaundice and the presence of IgM anti-HAV. Causality has not been universally accepted; those who maintain that direct evidence is lacking point out that the risk factors for certain groups of haemophiliacs were not always clearly stated and that in some cases epidemiological definition of the control groups mentioned was lacking (Robinson et al, 1992). Moreover, in 330 patients with haemophilia A studied in Norway, 28 of 202 in whom testing was performed had evidence of infection with HAV. However, it was shown in 27 patients (one emigrated) using archival samples that antibody was present before the institution of treatment with SD-treated FVIII (Evensen & Rollag, 1993). Other investigations in Finland and the USA have not revealed conclusive evidence of HAV transmission by SD-treated FVIII (Prowse et al, 1994). This should include an initial extensive investigation of a range of parameters on a relatively small number of units (for example 20 units of virally inactivated FFP compared with control FFP). This will involve in vitro studies, including those in which units are thawed and tested at intervals, e.g. 3, 6 and 12 months after manufacture, and more reliable data may be obtained where each subject donates twice (paired study design). In vivo studies in healthy volunteers may be required. One donation is handled according to standard procedures and the other is subjected to the viral inactivation (or other) strategy being tested before reinfusion. Currently, the guidelines for evaluation of new FFP/cryoprecipitate components for transfusion in the UK (Guidelines for Blood Transfusion Services in the United Kingdom, 2000) suggest the following. • Volume, platelet count, WBC (the last is particularly important where plasma filtration is included; in the UK, from November 1999, all FFP has been manufactured as leucodepleted). • Prothrombin time, partial thromboplastin time. • Specific coagulation factors assays for fibrinogen (FBG), factors II, V, VII, VIII, IX, X, XI, von Willebrand factor antigen (VWF:Ag), von Willebrand factor ristocetin cofactor activity (VWF:RiCof). • Analysis of VWF multimers on a small number of units only. • Inhibitors of coagulation – antithrombin III, protein C, protein S. • Markers of unwanted activation of coagulation – such as prothrombin fragment1+2 or fibrinopeptide A. • Markers of unwanted activation of kinin/complement – C3a, C5a, bradykinin, factor XIIa. Other assays such as fibrin split products and thrombin antithrombin (TAT) complexes may be helpful. Testing for the presence of neoantigens should also be carried out. It is also recommended that consideration is given to performing studies on representative units of FFP stored at −20°C as well as −30°C (to reflect differing hospital storage conditions). Sampling should be undertaken at 3, 6, 9 and 12 months. The minimum assays to be performed at each time point should include FBG and FV, VIII, IX, X and vWF:RiCof. It is also important to demonstrate that the agents used in viral inactivation systems such as methylene blue, solvents, detergents and psoralens are neither toxic nor mutagenic. These studies are usually undertaken by the manufacturer. Ideally, the systems in place should include their removal from plasma (≥ 99%) before it is frozen. Finally, it is important in model systems to demonstrate that significant antiviral activity (usually ≥ 4 log) exists against both lipid enveloped and, if possible, also non-lipid enveloped viruses. Testing may also assess antiviral activity against extracellular as well as intracellular viruses. The solvent–detergent treatment process was first licensed in 1985 for the manufacture of factor VIII concentrates (Horowitz & Ben-Hur, 1995). The treatment process had also been validated in the manufacture of antiviral vaccines and damages lipid but not proteins (Horowitz, 1991). The combination frequently used is 1% tri(n-butyl) phosphate (TNBP) in combination with a detergent, usually 1% Triton X-100. Pools of between 380 and 2500 ABO identical units of FFP are made after thawing (Horowitz & Ben-Hur, 1995). Contributing units may be either Rh D positive or negative. The SD treatment is performed at 30°C, usually for 4 h (Horowitz & Ben-Hur, 1995). The starting plasma is filtered to remove residual cells and cell debris and after treatment the TNBP is removed (< 2 µg/ml) using oil extraction and phase separation and Triton X-100 (< 5 µg/ml) by means of hydrophobic chromatography on C18 resin (Horowitz et al, 1992; Horowitz & Ben-Hur, 1995). It is claimed that the chromatographic step used to remove solvent is effective in removing HAV (Evensen & Rollag, 1993). The plasma is then sterile filtered (0·2 µm) and frozen in aliquots of 200 ml (Horowitz & Ben-Hur, 1995). The levels of coagulation factors in SD-FFP are generally equivalent to those in the start pool (Horowitz & Ben-Hur, 1995) and, indeed, recovery is usually > 90% (Table II) (Piët et al, 1990; Hellstern et al, 1992; Horowitz et al, 1992, 1998). Although levels of VWF:Ag and VWF:RiCof are > 90%, there is loss of some high molecular weight (HMW) VWF multimers during processing (Piquet et al, 1992). Factor VIII has been shown to be stable during 3 months of storage (Piquet et al, 1992). There is no evidence of activation of coagulation or fibrinolysis (Horowitz & Ben-Hur, 1995). Levels of plasminogen, C1 esterase inhibitor, TAT complexes and fibrin split products are normal but levels of α-2-antiplasmin and protein S are significantly reduced during the SD treatment process. Levels of d dimers are normal (Hellstern et al, 1992). No evidence of activation of coagulation factors has been reported (Piquet et al, 1992). A lack of neoantigenicity has been demonstrated in experiments in New Zealand white rabbits using crossed immunoelectrophoresis (Horowitz et al, 1992). A wide range of lipid-enveloped viruses are inactivated rapidly, usually within 15 min of addition of solvent and detergent (Horowitz et al, 1993; Biesert & Suhartono, 1998). A summary of viral inactivation studies is shown in Table III. In addition, it has been shown that an SD-treated immunoglobulin prepared from patients with HIV did not transmit infection to chimpanzees (Hellstern et al, 1992). Some viruses are studied because they cause clinical infection in man, whereas others act as models for viruses which cannot themselves be grown in culture, e.g. pseudorabies and duck HBV (DHBV) are model viruses for HBV and bovine viral diarrhoea virus (BVDV) is a model virus for hepatitis C. SD treatment does not inactivate non-lipid-encapsulated viruses such as HAV and parvovirus B-19. This has led to concerns that overall benefit might be limited as non-irradiated viruses in a batch of SD-FFP could be transmitted to a large number of susceptible recipients (Williamson & Allain, 1995). Partitioning of the immunoglobulin fraction during preparation of coagulation factor concentrates leads to loss of protective IgG antibody, but in the case of FFP it has been estimated that pooling of donated plasma results in levels of IgG anti-HAV 30 times higher than that in intramuscular immunoglobulin preparations, which have not been shown to transmit hepatitis. Levels of antiparvovirus B-19 IgG antibody are similar to those found in IVIG that can be used for treatment of persistent parvovirus infections (Horowitz & Ben-Hur, 1995; Horowitz et al, 1998) and, in data from Norway, showed uniform virus neutralization (Solheim et al 2000) (see below). However, in the USA, 19 batches of SD-FFP have been recalled because of seroconversions to parvovirus B-19 and detection of B-19 DNA by PCR in volunteers who had received SD-FFP (CBER Document no. 0694, 1999). The former might have been due to passive transfer of antibody. In addition, batch testing revealed higher than acceptable levels of parvovirus B-19 nucleic acid. It has been shown that by screening donor blood by genomic testing (PCR) for HAV and parvovirus B-19, SD-FFP can be prepared from PCR-negative plasma pools (Horowitz et al, 1999). The clinical sequelae of B-19 infection include marrow aplasia in patients with haemolytic anaemias or other erythropoietic abnormalities and hydrops fetalis if transfused during pregnancy. SD-FFP has not, to date, been associated with documented viral transmission in clinical practice. In a study of 343 adults undergoing cardiac surgery, 194 received transfusion support which included SD-FFP and 41 patients received only SD-FFP. Twenty-five batches of SD-FFP prepared from Norwegian plasma were used. All batches contained neutralizing antibodies to B-19 and 20 of 25 to HAV, whereas in the other five the anti-HAV reactivity was borderline. After a 6- to 12-month follow-up period to allow washout of transfused antibodies, testing revealed seroconversion to HAV in four patients, to HBV (anti-HBc) in one patient, to CMV in four patients and to parvovirus B-19 in nine patients. No seroconversions could be ascribed to transfusion of SD-FFP (Solheim et al 2000). Cryoprecipitate has been prepared from SD-treated pooled FFP. In SD cryoprecipitate, levels of VWF activity and antigen were reduced to only 36% and 30% of control values respectively. The fibrinogen level was 72%. The highest molecular weight multimers of VWF were absent from cryoprecipitate, indicating that this product could be an alternative to standard cryoprecipitate for the treatment of hypofibrinogenaemic states but unsuitable for treating von Willebrand's disease (Keeling et al, 1997). There has been no evidence of toxicity when SD-treated blood components or fractionated products are transfused. In a haemovigilance study in Belgium, no adverse events were reported during a year when 5064 units of SD-FFP were transfused to 894 patients (Baudoux et al, 1998). Occasionally, FFP from a different donor is used for the resuspension of either red cells or platelets before transfusion. Patients who require massive transfusion, for example after trauma, receive large volumes of blood components, including red cells, platelets and FFP. Studies over 5 d of storage using both red cells and platelets have shown no adverse effects of SD-FFP on parameters of red cell or platelet function (Tocci et al, 1993; Snyder et al, 1994). Studies using SD-FFP in volunteer subjects were not conducted before its clinical evaluation in patients. Subsequently, phase IV studies have been carried out in healthy volunteers (CBER Document no. 0694, 1999) but data concerning clinical efficacy, for example coagulation factor recovery, are not published at the present time. Coagulopathy In 11 patients with hereditary deficiency of FVII, X or XI, administration of SD-FFP resulted in cessation or prevention of haemorrhage; the factor half-lives were as expected (Inbal et al, 1993). Similar findings were reported in patients with FXIII deficiency during 39 episodes of prophylaxis (Horowitz & Pehta, 1998). In the UK, haemophilia directors have recommended the use of virally inactivated FFP to treat single-factor deficiencies when no factor concentrate is available (United Kingdom Haemophilia Centre Directors Organization, 1997). A significant improvement in the prothrombin time and FBG level was demonstrated together with stabilization of blood pressure and cessation of clinical bleeding in 16 of 22 patients with disseminated intravascular coagulation (DIC) (Hellstern et al, 1993). Reports describe the use of SD-FFP in 75 patients with surgical bleeding and nine patients in whom warfarin reversal was required preoperatively (Horowitz & Ben-Hur, 1995). In addition, a group of 48 patients with coagulation factor deficiencies received a total of 788 units of FFP. These were given for surgical prophylaxis (47 episodes), active bleeding (51 episodes) and in FXIII deficiency (see above) (Horowitz & Pehta, 1998). In both reports, the expected levels of coagulation factors were achieved and bleeding was adequately controlled. Heart surgery In 122 patients of whom 53 (43%) required plasma, SD-FFP was offered to 46 patients and given to 20. Twenty controls received standard FFP. There were no perioperative differences in cardiovascular or respiratory support given or transfusion requirements. Correction of the prothrombin time was equivalent and there was no laboratory evidence of complement activation (Solheim et al, 1993). Thrombotic thrombocytopenic purpura (TTP) In some patients with this condition, abnormally large high molecular weight (HMW) VWF multimers are present. Cryosupernatant plasma lacks large VWF multimers and could be advantageous in treatment. A recent report suggests at least equivalent, and possibly superior, efficacy compared with standard FFP (Rock et al, 1996). As SD-FFP lacks some HMW VWF multimers, it may also be particularly useful in this group of patients. A number of reports of plasma infusion or exchange using SD-FFP show no adverse effects during treatment (Moake et al, 1994; Pehta, 1996; Evans et al, 1999). Two out of five patients with chronic relapsing TTP received alternate treatments with conventional vs. SD-FFP and 3/5 received only SD-FFP on study, with their responses compared with their historical control data; responses in platelet count and duration of remission were similar and no adverse events were reported. In a further two children with chronic relapsing TTP who received SD-FFP infusions, evidence of VWF-mediated, sheer stress-induced platelet aggregation together with an increase in HMW VWF multimers in platelet-poor plasma was observed. After plasma infusion, abnormal HMW VWF multimers disappeared within 1 h, reversal of abnormal sheer stress in 8/9 infusions studied occurred within 1–4 h and the platelet count usually normalized after a week. Therefore, an important plasma constituent responsible for the control of TTP is unaffected by the SD process (Moake et al, 1994). A further six patients with chronic relapsing TTP who received SD-FFP infusions showed a good response in the platelet count with reduction of lactate dehydrogenase (LDH) and stabilization of the haemoglobin (Horowitz & Pehta, 1998). In three patients with acute TTP all treated using plasma exchange against SD-FFP, the platelet count rose to > 50 × 109/l by day 3 in one patient, by day 7 in another patient and by day 10 in the third patient. All remain in remission > 1 year later (Evans et al, 1999). Liver disease In a multicentre UK study, 24 liver disease and 25 liver transplant patients were randomized to received SD-treated or standard FFP. Equivalent correction of the international normalized ratio (INR), activated partial thromboplastin time (APTT) ratio and levels of FII, VII and protein C were noted. Blood component usage in each group for liver transplant patients was equivalent. In at-risk patients with sufficient follow-up, no seroconversions to HBV or HCV were seen and although seroconversion to HAV was suspected originally in four patients this was thought to be the result of transfused IgG antibody as three out of four had no detectable IgM anti-HAV and the IgG antibody eventually became undetectable. HAV-RNA testing was not carried out on these patients (Williamson et al, 1999a). Children Few data on the use of SD-FFP are available in children. In two neonates and 18 children (1·5–17 years), no untoward effects followed SD-FFP transfusion and in Norway 75 transfusions of SD-FFP were given uneventfully to neonates and older children (Klein et al, 1998). Front-end filtration of FFP before SD treatment removes cell and cellular/particulate debris. Leucocyte depletion reduces both the incidence of alloimmunization (AI) to HLA and the occurrence of non-haemolytic febrile transfusion reactions (NHFTR). Currently, in the UK, universal leucodepletion of all blood components has been introduced to reduce the theoretical risk of transfusion transmission of the pathogenic agent of new variant Creutzfeld-Jakob disease (nvCJD). However, studies using a time-resolved fluoroimmunoassay for the cellular isoform of prion protein (PrPc) in human blood has shown that 68·5% is associated with plasma and 26·5% with platelets (MacGregor et al, 1999). The significance of this regarding acquisition of the causative agent of nvCJD is unknown at the present time. Terminal filtration using a 0·2-µl filter effectively removes bacteria (Horowitz & Ben-Hur, 1995). Theoretically, the presence of neutralizing antibody to pathogenic non-enveloped viruses should prevent their transmission, although it is possible that this could occur if antibody in the donor population was present at a low frequency or was non-neutralizing. Because of the extreme dilution that occurs when pooling, SD-FFP may be beneficial in reducing the number of reactions that occur as a result of infusion of antibodies present in high titre in individual donations, for example transfusion-related acute lung injury (TRALI), and allergic reactions, due to dilution when pooling plasma. As red cells are removed, Rh sensitization does not occur and it is not necessary to state the Rh type of SD-FFP. A recent study reported a net benefit of 35 min of quality-adjusted life expectancy costing an extra $19 for the viral inactivation process. This would equate to $289 300 per quality-adjusted life year (QUALY) saved. The authors of this analysis concluded that the expenditure was unjustified (AuBuchon & Birkmeyer, 1994). A more recent study taking account of declining virus transmission rate and more recent knowledge of the costs of viral inactivation procedures concluded that transfusing virally inactivated FFP prolonged quality-adjusted survival by 1 h 11 min per patient, at a cost-effectiveness ratio of $2 156 398 ± $257 587 per QUALY (Periera, 1999). It is generally agreed that QUALY costs of more than $50 000 are not cost-effective. The use of methylene blue for viral inactivation of plasma was first described in 1991. It is a phenothiazine dye previously used in the treatment of methaemoglobinaemia and shown to inactivate viruses in media, plasma and red cells. It binds to DNA but also has an affinity for the surface structures of viruses (Lambrecht et al, 1991). Methylene blue in combination with visible light (red/white) in the presence of oxygen causes a formation of 8-hydroxyguanine and DNA–protein cross-links (Lambrecht et al, 1991; Wagner et al, 1998). The DNA affinity of phenothiazines varies. In a comparative study of dimethylmethylene blue (DMMB), methylene blue and two closely related analogues designated 6-136 and 4-140, DMMB had the highest DNA binding and was able to kill the non-encapsulated bacteriophage R17. In contrast, the least active compound 4-140, a hydrophobic molecule with high protein affinity, effectively killed a lipid-encapsulated bacteriophage, illustrating that whereas phenothiazines exert their effect primarily via DNA damage they also interact with the viral envelope, although in this case the virucidal effect was inhibited by red cells and plasma (Lambrecht et al, 1991). Methylene blue is usually added to plasma to a final concentration of 1 µm. As methylene blue is inactive against intracellular viruses, plasma is frozen and thawed first to disrupt leucocytes (Mohr et al, 1997). Alternatively, leucocytes can be removed before freezing by filtration either of whole blood before FFP separation or of secondarily separated FFP. In a comparison, the mean residual viable leucocyte counts of freeze-thawed and filtered plasma were 0·06 × 106/l and < 0·002 × 106/l respectively (Rider et al, 1998). After leucocyte removal and addition of methylene blue, units of plasma are exposed to white light for approximately 30 min to 1 h. Until recently, the process did not involv